430 
Fishery Bulletin 106(4) 
ments on Pacific halibut were conducted approximately 
10-12 hours after exposure to bright light. 
Three separate procedures were conducted to test 
for treatment effects on visual function: responses to 
increasing light intensities (V-log I response curves), 
flicker fusion frequency, and spectral sensitivity. For 
the first two procedures, light stimuli were produced 
by a circular (3.8 cm diameter) light source (model 
SL2420, Advanced Illumination, Rochester, VT) com- 
prising 20 white LEDs. The LEDs were mounted behind 
a thin diffuser and collimating lens to produce an even 
(±10% edge to edge) illumination field. Light output 
was controlled by an intensity controller (model CS410, 
Advanced Illumination, Rochester, VT), which in turn 
was controlled the analog output of the data acquisition 
card. A series of neutral density filters (Kodak Optical 
Products, Rochester, NY) were used to extend the range 
of light levels as needed. 
For the determination of spectral sensitivity, the 
output of a xenon fiberoptic light source (model Y1603, 
CVI Laser Spectral Products, Albuquerque, NM) was 
controlled with a monochrometer (model CM110, CVI 
Laser Spectral Products, Albuquerque, NM), two filter 
wheel assemblies (model AB301, CVI Laser Spectral 
Products, Albuquerque, NM) containing quartz neu- 
tral density filters, and an electronic shutter (model 
LS6, Uniblitz, Vincent Associates, Rochester, NY). 
The monochrometer produced single wavelength light 
(with an 8 nm 50% bandwidth) between 300 and 700 
nm. The filter wheel assemblies allowed the attenua- 
tion of light from 0 to 5 log units in 0.2 log-unit steps. 
The monochrometer and filter wheel assemblies were 
controlled by using serial (RS232) interfaces, and the 
shutter was controlled by the digital output of the data 
acquisition card. Light from the xenon light source 
was conducted through the monochrometer — >- filter 
wheel — > shutter assembly, and from there directed 
at the eye through 3-mm light guides. The LED light 
source and exit point of the light guide were placed 
approximately 5 cm from the corneal surface. The 
outputs of the LED and xenon light sources (the lat- 
ter measured at exit of the light guide after passage 
through the monochrometer — > filter wheel —>■ shutter 
assembly) were calibrated with a research radiometer 
(model IL 1700, International Light, Inc., Newbury- 
port, MA). 
To construct V-log I response curves, light intensities 
were increased in 0.2 log-unit steps from levels that 
produced no measurable responses, to those that pro- 
duced maximal responses. Stimuli consisted of a train 
of five 200-ms duration light flashes 200 ms apart. The 
trains of stimuli at each light intensity were presented 
every five seconds and repeated five times. The ERG 
responses to the final flash of each train were recorded 
and averaged. The data were subsequently normalized 
by expressing the average response to an intensity 
step as a fraction of the maximum observed average 
response. Normalized ERG responses versus log light 
intensities (candela/meter 2 ) were plotted to construct 
V-log I response curves. Mean V-log I curves for each 
species were constructed by averaging the normalized 
curves of all individuals within a treatment group re- 
corded during the day or during the night. 
Flicker fusion frequencies were determined by using 
five-second sinusoidal light stimulus trains, followed by 
five seconds of darkness. The maximum light intensity 
of the sinusoidal stimulus was that required to produce 
a response that was 50% of the maximal response. This 
value was determined by eye during experiments from 
the individual V-log I response curves. Stimulus trains 
were repeated five times at each frequency and the re- 
sponses averaged. Stimulus frequencies were increased 
from 1 Hz (0 log units) to 63.1 Hz (1.8 log units) in 0.2 
log-unit steps. The flicker fusion frequency was deter- 
mined by comparing the power spectrum of the aver- 
aged response at each stimulus frequency (signal) to 
the power spectrum of a neighboring frequency (noise) 
and was defined as the frequency at which the power 
of the signal fell below the power of the noise. 
Spectral response curves were determined using 
monochromatic light flashes produced by a xenon light 
source and monochrometer —> filter wheel — > shutter 
assembly. Approximately isoquantal light stimuli from 
300 to 650 nm, regulated by the monochrometer and 
neutral density filters, were presented to subjects in 
10-nm wavelength steps. Five stimuli of 40 ms duration 
were presented at each wavelength, and five seconds 
were allowed between each light flash. The responses 
to the five flashes were averaged and mean amplitudes 
recorded. 
Spectral sensitivity curves were subsequently cal- 
culated from the spectral response curves as follows. 
First, spectral response data were corrected for dif- 
ferences in lamp output at specific wavelengths, as 
well as for differences in neutral density filter values 
from their nominal values, through the application of 
correction factors. The correction factors were based 
on the calibration curves developed with the research 
radiometer described above. Second, responses to ex- 
actly isoquantal intensities were predicted by adjusting 
the response amplitude at each wavelength by using 
spectral V-log I response curves generally following 
the methods described in Coates et al. (2006). The only 
exception was that a fourth-order polynomial was fitted 
to the individual spectral V-log I response data. Spec- 
tral V-log I response curves were recorded immediately 
after measurement of the spectral response curves 
by using a series of increasingly intense flashes at 
the wavelength which generated the largest response. 
Five flashes (200 ms duration, five seconds apart) were 
delivered at each intensity step and the response am- 
plitudes averaged. Intensities increased in 0.2 log-unit 
steps and ranged from those producing no response, to 
those producing a maximal response. 
Individual spectral sensitivity curves were normal- 
ized by using the maximal response at each wavelength. 
Mean spectral sensitivity curves were constructed by 
averaging the spectral sensitivity curves of all individu- 
als within a treatment group recorded during the day 
or during the night 
